|Publication number||US4839724 A|
|Application number||US 07/125,497|
|Publication date||Jun 13, 1989|
|Filing date||Feb 7, 1987|
|Priority date||Feb 18, 1986|
|Also published as||DE3605032A1, EP0258341A1, EP0258341B1, WO1987005178A1|
|Publication number||07125497, 125497, PCT/1987/60, PCT/EP/1987/000060, PCT/EP/1987/00060, PCT/EP/87/000060, PCT/EP/87/00060, PCT/EP1987/000060, PCT/EP1987/00060, PCT/EP1987000060, PCT/EP198700060, PCT/EP87/000060, PCT/EP87/00060, PCT/EP87000060, PCT/EP8700060, US 4839724 A, US 4839724A, US-A-4839724, US4839724 A, US4839724A|
|Inventors||Heinz-Werner Keesen, Hartmut Peters|
|Original Assignee||Deutsche AG fur Unterhaltungselectronic OHG|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (8), Non-Patent Citations (4), Referenced by (22), Classifications (19), Legal Events (5)|
|External Links: USPTO, USPTO Assignment, Espacenet|
The invention is concerned with a process for the digital transmission of information especially for digital video signals obtained fromm analog signals, in which groups of data are processed by a source coder and are sent via a channel coder to a transmission channel and from the transmission channel via a channel decoder to a source decoder.
European Patent Application Publication No. 0,084,270 corresponding to U.S. Ser. No. 336,984 of Apr. 1st, 1982, now U.S. Pat. No. 4,541,012 discloses a process for the encoding of transformed coefficients. A plurality of code tables are used to generate a block of address codes, a DC-coefficient code, and a sequence of codes for transformed coefficients.
The invention is based on the task to improve transmission processes and/or to optimize the information content of a signal or information, respectively.
The task is solved, in general according to the present invention by encoding the signals from the source coder according to a variable codeword length code, and adapting the variable cordeword length code, at time intervals, according to a statistical value of the signal values actually being processed and transmitted.
FIG. 1 is a block circuit diagram of a basic known transmission system for use with the method according to the invention;
FIG. 2 is a more detailed block diagram of a source coder for use in the circuit of FIG. 1;
FIG. 3 shows a coefficient block or matrix whose values are to be transmitted;
FIG. 4 is a table showing the relationship between the values of FIG. 3 and a Huffman coding process;
FIG. 5 shows a Huffman coding arrangment;
FIG. 6 shows a zigzag scan for the transmission of the coefficients of FIG. 3;
FIGS. 7A and 7B each show a block of coefficients with zigzag-scan sampling and the resulting Huffman code according to the table of FIG. 4;
FIG. 8 shows a transmission quantities according to the preferred embodiment of the invention;
FIG. 9 shows a display screen;
FIG. 10 is a block circuit diagram of an embodiment of a transmitter for use in the method according to the invention;
FIG. 11 is a block circuit diagram of an embodiment of a receiver for use in the method according to the invention;
FIG. 12 is a more detailed block circuit diagram of a portion of a transmitter for use with the method according to the invention and particularly a preferred embodiment of the Huffman coder arrangement;
FIG. 13 is a more detailed block circuit diagram of a portion of a receiver for use with the method according to the invention and particularly a preferred embodiment of the Huffman decoder.
FIG. 1 shows a transmission system including an information source 1, an ADC 2 (analog-digital converter), a source coder 3, a channel coder 4, a channel 5, a channel decoder 6, a source decoder 7, a DAC 8 (digital-analog converter) and an information sink 9. The channel 5 can, for instance, represent a HF transmission line or a digital videorecorder. A digital videorecorder, line or a digital videorecorder. A digital videorecorder, however, can also represent the ADC 2, the source coder 3, the channel coder 4, the channel 5, the channel decoder 6, the source decoder 7 and the DAC 8. The source 1 is then formed by a HF transmission line, the channel 5 by a magnetic tape and a magnetic head of the videorecorder, and the sink by a television set.
FIG. 2 shows that in the source coder 3, a digital signal runs, firstly, through a ADCT 10 (adaptive discrete cosine transformer), secondly, through a weighting member 11, thirdly, through a block quantizer 12, fourthly, through a coder 13 and, fifthly, through a buffer 14. Especially if images are sent to a source coder 3 without intermediate storage, the block quantizer 12 must be controlled so that, at the receiving end, sufficient data for the production of an image with acceptable image quality are always available.
FIG. 3 shows a block of distribution coefficients as it arrives at the block quantizer 12 after an adaptive discrete cosine transformation. The point marked with X contains the distribution coefficient for the steady component of a matrix, called the mean value hereinafter. In practice, a 8 by 8 matrix is used, corresponding to 64 distribution coefficients and 64 pixels, respectively. The 8 by 8 matrix or block contains 64 distribution values or spectral coefficients 0, 1, 2, . . . , 7, called the values hereinafter.
FIG. 4 shows a table for the values 0, 1, 2, . . . , 7. The eight values can be represented in binary code by means of three binary symbols for each value. According to the Huffman coding technique, binary symbols can be saved during a transmission. That is, the frequency of occurrence of the values to be transmitted is determined, and the most frequently transmitted value is supplied with fewer binary symbols than the least frequently transmitted value. The value 0 appears nineteen times in the 8 by 8 block, the value 1 twelve times, the value 2 eight times, and so on. Following the Huffman code, the code 01 is transmitted for the value 0, the code 11 for the value 1, and the code 1011 for the value 7.
FIG. 5 shows the calculation scheme according to the Huffman code. Here the frequencies are listed in vertical sequence. The scheme is based on a tree or branching pattern. The lowest frequencies in each case are collected in blocks of two and are added up. Furthermore, sums and/or lowest the frequencies are again added in blocks of two. The collection is done so that the lower values arrive at the sum in a vertical line whereas the higher values do so in a horizontal line. Horizontal lines have the value 0, vertical lines the value 1. The Huffman code is also explained on Pages 52-55 of Robert C. Gallagher, Information Theory and Reliable Communication, copyright 1968 by John Wiley & Sons, Inc. ISBN W-471-29048-3, Library of Congress Catalog Card No. 68-26850.
FIG. 6 shows a zigzag scan according to EP-OS No. 0,084,270 European Patent Application No. 0,084,270 and corresponding to U.S. Pat. No. 4,541,012. The scan indicates the sequence of the values to be transmitted.
FIG. 7A shows values 0310, which are to be transmitted according to the Huffman code and the zigzag scan. This results in the binary sequence 011001101.
FIG. 7B shows values 6766, which are to be transmitted according to the zigzag scan and the Huffman code. This results in the binary sequence 1010101110101010. The binary sequence for the values 6766 has more binary symbols to be transmitted in comparison with the binary sequence 0310. If the frequency of the values to be transmitted is checked at regular time intervals, the frequency may be different and a new Huffman code may be prepared.
FIG. 8 shows quantities or blocks 15-19, in which the information to be transmitted is listed. Transmitted are the mean value (15), the number of the distribution coefficients (16), the amount of the distribution coefficients (17), the transmission of the signs for distribution coefficients (18), and the addresses, control symbols, and sync words (19). Advantageously, the sequence of the quantities to be transmitted is defined as follows: 19, 15, 16, 17, 18.
FIG. 9 shows a television receiver 20 with time-domain blocks 21 that each contain 64 pixels. The time-domain blocks 21 are transformed, block by block, into the spectral domain. This is done, for instance, by means of the ADCT 10.
FIG. 10 shows the reduction of values to be transmitted with an aid of the adaptive Huffman coder according to the preferred embodiment of the invention. For this purpose, a Huffman coder 22 is connected between a source coder 3 and a channel coder 4. In the actual code book 24, the assignments of the values and the Huffman code are stored. If a value appears in the Huffman coder 22, arriving from the source coder 3 via the control line 26, the Huffman code for the value is called up over the control line 33. At the beginning of a transmission, the actual code book 24 is loaded with the contact of a standard code book 23 via the control 25. A line 30 is disposed between standard code book 23 and control 25. A bidirectional line 32 lies between control 25 and actual code book 24. If the actual code book 24 is adapted to the transmitted values, a change occurs between the actual code book and the standard code book 23. The control 25 examines the changes between the actual code book 24 and standard code book 23 and communicates this change to the channel coder 4 via the line 29. For this purpose, the Huffman coder is switched off for this interval via the line 31. Advantageously, a synchronization word, called sync word hereinafter, precedes this communication of change. By means of the channel coder 4, this signal can be recorded on the magnetic tape of a videorecorder, or the signal is transmitted via an HF transmission line. During the switching-on of a sender, the actual code book 24 is loaded with the assignments of the standard code book 23. The control 25 interrogates the standard code book 23 and loads the assignments into the actual code book 24 via the bidirectional line 32. The Huffman coder 2 sends the Huffman-coded values to the channel coder 4 via a line 27 and the nodal point 28. Communications of changes are possible on the basis of a previous actual code book 24 and a new actual code book 44. Advantageously, communications of changes are referenced only to the standard code book 23.
FIG. 11 shows a channel decoder 6, which receives a signal from the magnetic head of a videorecorder or from a HF transmission line. The signals are decoded in the decoder 16 and reach the computer 35 and the Huffman decoder 34 via the lines 36, 37, and 38. During the switching-on of the receiver, the computer 35 loads the actual code book 44 via the bidirectional line 39 with a standard coding regulation from the standard code book 23. The standard coding regulation is defined and is identical for receiver and transmitter. A line 30 lies between standard code book 23 and computer 35. The Huffman decoder 34 interrogates the actual code book 44 via the bidirectional line 33 when signals arrive via the line 38 and assigns the values to the incoming Huffman code. Values are passed on to the source decoder 7 via the line 41. If the computer 35 recognizes control symbols over the line 37, the Huffman decoder 34 is switched off via the line 38 and the actual code book 44 is updated or adapted on the basis of the sync words, control signals, and communications of changes.
FIG. 12 shows a Huffman coding arrangement according to the invention. A shift register 54 is loaded with values via the line 26. The values have a length of 8 bits. The 8 bits are used as an address and drive an address decoding member 42 by means of an address bus 56. By means of eight addresses, 256 memory locations, that is to say, 28 can be addressed. The Huffman code is in the memory 58. A maximum codeword length of 11 bits can be stored in the memory 58. Via the memory 59, the codeword length can be called up. The codeword length is coded by means of 3-bit line. The memories 60 and 61 correspond to the memories 58 and 59. Two memory units SP1, SP2, (58, 59) and SP3, SP4, (60, 61) are necessary to update the actual code book 24. The memories 58-61 and the address decoding member 42 corresponding to the actual code book 24 of FIG. 10 and may be configured, for instance, by four RAM TMS 4416-15. The standard code book 23 may be configured, for instance, by two EPROM TMS 2516. Eleven assigned values can be recalled via 8 addresses. For this see Texas Instruments, MOS Memory Data Book, 1984--European Edition. In the control 25, the number of values of the same magnitude is counted over a predetermined interval of time. Actualization, i.e., adaptation or updating of the actual code book 24, can be done according to the number of values. If the previous actual code book is in the memories 58, 59, the new actual code book is written into the memories 60 and 61. After an actualization, the actual position is communicated in the shift register 52 via the control line 29 by the control 25. For this purpose, the inputs, at the shift register 52, of the lines between the shift register 52 and the memories 58-61 are blocked. The shift register 52 transmits, during the switch-off, sync words and the actualization, i.e., the new actual code book, to the channel coder 4 for transmission.
It is advantageous firstly, to make an adaptation of the Huffman code in the actual code book 24 on the basis of a threshold-value decision, secondly, to add the number of values over a period of two television images and, after a time period of two television images, undertake an adaptation of the Huffman code, or, thirdly, to make a combination of both adaptations. During a threshold-value decision, the number of values is added up and the adaptation is made after crossing of a predetermined threshold.
FIG. 13 shows a Huffman decoder. The computer 35 recognizes sync words via the line 37 and thus a change of the actual code book. Huffman-coded signals are given to the shift register 43 via the line 38. A length of 11 bits may appear in the shift register 43 after the Huffman coding. The Huffman coding of 11 bits corresponds to an address space of 256 bits, that is to say, 28. The address decoding member 49 has an address space of 211, that is to say, 2048 bits. Of these addresses, 256 are for Huffman decoding. The remaining memory locations indicate that no valid Huffman codeword is in the shift register. The memories 45 and 46 correspond to the memories 47 and 48. The shift register 43 is loaded bit by bit. The memories 45-48 and the address decoding member 49 correspond to the actual code book 44 of FIG. 11. All bits are set at zero in the shift register 43. When the Huffman-coded values arrive via the line 38, the bits in the shift register 43 are serially shifted. In each case, upon arrival of a bit, a check is made in the address decoding member 49 whether, according to value in the shift register 43, an address assignment is possible. In this case, 2, 3, 4, . . . or 11 bits could have arrived in the shift register 43. If a Huffman coding is recognized, the assigned value is transmitted from the memory 45 to the shift register 62. At the same time, the shift register 43 is set back to the original state, i.e., all bit points are loaded with zero, by the memory 46 via the OR member 50.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US4292651 *||Feb 9, 1979||Sep 29, 1981||Francis Kretz||Expansion and compression of television signals by use of differential coding|
|US4302775 *||Dec 15, 1978||Nov 24, 1981||Compression Labs, Inc.||Digital video compression system and methods utilizing scene adaptive coding with rate buffer feedback|
|US4394774 *||Jun 26, 1981||Jul 19, 1983||Compression Labs, Inc.||Digital video compression system and methods utilizing scene adaptive coding with rate buffer feedback|
|US4541012 *||Jan 4, 1982||Sep 10, 1985||Compression Labs, Inc.||Video bandwidth reduction system employing interframe block differencing and transform domain coding|
|US4586182 *||Feb 6, 1984||Apr 29, 1986||Codex Corporation||Source coded modulation system|
|US4694336 *||Apr 14, 1986||Sep 15, 1987||Telefunken Fernseh Und Rundfunk Gmbh||Digital data transmission method|
|US4706264 *||Jun 22, 1983||Nov 10, 1987||Chung Telecommunications||Digital data compression method and means|
|US4707738 *||Dec 23, 1985||Nov 17, 1987||Thomson Grand Public||Adaptive process for the coding and decoding of a sequence of pictures by transformation and devices for performing this process|
|1||Dukhovich; "A DPCM System Based on a Composite Image Model"; National Telecommunications Conference; Band 2, vol. 4; pp. 36.2.1-36.2.6; Nov. 30-Dec. 4, 1980.|
|2||*||Dukhovich; A DPCM System Based on a Composite Image Model ; National Telecommunications Conference; Band 2, vol. 4; pp. 36.2.1 36.2.6; Nov. 30 Dec. 4, 1980.|
|3||W. Chen; "Scene Adaptive Coder"; International Conference on Communication; vol. 2, pp. 22.5.1-22.5.6; Jun. 14-18, 1981.|
|4||*||W. Chen; Scene Adaptive Coder ; International Conference on Communication; vol. 2, pp. 22.5.1 22.5.6; Jun. 14 18, 1981.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US5058137 *||Jul 31, 1989||Oct 15, 1991||North American Philips Corporation||Lempel-Ziv decoder|
|US5325124 *||Dec 22, 1992||Jun 28, 1994||Intel Corporation||Method and apparatus for scalable decompression of a digital motion video signal using variable pixel depths|
|US5363139 *||Dec 22, 1992||Nov 8, 1994||Intel Corporation||Method and apparatus for scalable compression and decompression of a digital motion video signal|
|US5420639 *||Apr 1, 1993||May 30, 1995||Scientific-Atlanta, Inc.||Rate adaptive huffman coding|
|US5455874 *||Sep 20, 1993||Oct 3, 1995||The Analytic Sciences Corporation||Continuous-tone image compression|
|US5530478 *||Aug 21, 1992||Jun 25, 1996||Kabushiki Kaisha Toshiba||Image data compressing apparatus|
|US5577190 *||Jul 5, 1994||Nov 19, 1996||Avid Technology, Inc.||Media editing system with adjustable source material compression|
|US5786780 *||May 27, 1997||Jul 28, 1998||Samsung Electronics Co., Ltd.||Variable-length encoder and decoder using symbol/code-word re-association of a coding table|
|US5793432 *||Aug 29, 1997||Aug 11, 1998||Mitsubishi Denki Kabushiki Kaisha||Encoder and decoder|
|US5793896 *||Mar 23, 1995||Aug 11, 1998||Intel Corporation||Ordering corrector for variable length codes|
|US5990963 *||Jun 9, 1998||Nov 23, 1999||Mitsubishi Denki Kabushiki Kaisha||Encoder and decoder|
|US6023531 *||Jun 24, 1997||Feb 8, 2000||Avid Technology, Inc.||Quantization table adjustment|
|US6118444 *||Jul 10, 1996||Sep 12, 2000||Avid Technology, Inc.||Media composition system with enhanced user interface features|
|US6169820||Sep 12, 1995||Jan 2, 2001||Tecomac Ag||Data compression process and system for compressing data|
|US6553142||Jul 17, 2002||Apr 22, 2003||Avid Technology, Inc.||Quantization table adjustment|
|US6563875 *||Oct 17, 2001||May 13, 2003||Thomson Licensing S.A.||Adaptive method of encoding and decoding a series of pictures by transformation, and devices for implementing this method|
|US6687407||Apr 22, 2003||Feb 3, 2004||Avid Technology, Inc.||Quantization table adjustment|
|US8472624 *||Dec 10, 2010||Jun 25, 2013||Irdeto B.V.||Generating a scrambled data stream|
|US20110317833 *||Dec 10, 2010||Dec 29, 2011||Irdeto B.V.||Generating a scrambled data stream|
|USRE37912 *||Jul 28, 2000||Nov 26, 2002||Samsung Electronics Co., Ltd.||Variable-length encoder and decoder using symbol/code-word re-association of a coding table|
|WO1996008872A1 *||Sep 12, 1995||Mar 21, 1996||Tecomac Ag||Data compressing process and system|
|WO1996008892A1 *||Sep 12, 1995||Mar 21, 1996||Videocom, Inc.||Video distribution system using in-wall wiring|
|U.S. Classification||375/240.12, 370/472, 375/E07.266, 375/253, 375/E07.244, 382/239|
|International Classification||H04N19/593, H04N19/152, H03M7/42, H04N1/41, H04L23/00, H04B14/04|
|Cooperative Classification||H04N19/50, H04N19/593, H03M7/42, H04N19/152|
|European Classification||H04N7/34B, H03M7/42, H04N7/32B|
|Oct 13, 1987||AS||Assignment|
Owner name: DEUTSCHE AG, FUR UNTERHALTUNGSELECTRONIC OHG, D-77
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST.;ASSIGNORS:KEESEN, HEINZ-WERNER;PETERS, HARTMUT;REEL/FRAME:004800/0561
Effective date: 19870810
Owner name: DEUTSCHE AG, FUR UNTERHALTUNGSELECTRONIC OHG, D-77
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:KEESEN, HEINZ-WERNER;PETERS, HARTMUT;REEL/FRAME:004800/0561
Effective date: 19870810
|Dec 4, 1992||FPAY||Fee payment|
Year of fee payment: 4
|Nov 20, 1996||FPAY||Fee payment|
Year of fee payment: 8
|Jan 21, 1997||REMI||Maintenance fee reminder mailed|
|Nov 24, 2000||FPAY||Fee payment|
Year of fee payment: 12